opencli-adapter-author

Use when writing an OpenCLI adapter for a new site or adding a new command to an existing site. Guides end-to-end from first recon through field decoding, ad...

This skill is a detailed, coherent authoring guide and does what it says: it will direct an agent to use your opencli CLI and browser tooling, read and write files under ~/.opencli, create adapter files under clis/<site>/, and (via the browser tool) read cookie jars (including HttpOnly cookies via CDP). Before using it: (1) be prepared to review any file changes/commits the agent makes in your repo; (2) do not let the agent persist real auth tokens or raw cookies into repo files or fixtures — the guide reminds you to sanitize fixtures but that is not enforced; (3) run the agent in an environment with limited privileges or a throwaway workspace if you are unsure; (4) validate any injected interceptor code or helper snippets before allowing them to run; (5) confirm the opencli tooling invoked is the legitimate binary you expect. If you follow those precautions, the skill appears internally consistent and appropriate for its purpose.

Type: OpenClaw Skill Name: opencli-adapter-author Version: 1.0.0 The skill bundle provides a comprehensive framework for an AI agent to author OpenCLI adapters for various websites. It includes detailed runbooks, reference templates, and site-specific memory (e.g., for Bilibili and Eastmoney) to guide the agent through reconnaissance, API discovery, and field decoding. While the instructions involve high-risk operations such as extracting browser cookies (page.getCookies) and executing JavaScript in a browser context (opencli browser eval), these actions are strictly aligned with the tool's stated purpose of automating web data extraction. There is no evidence of intentional data exfiltration, malicious persistence, or prompt injection designed to subvert the agent's safety protocols.